State of the Art on Self-adaptive Systems: An Essay
Sara Mahdavi Hezavehi, Danny Weyns, Paris Avgeriou
TL;DR
The paper addresses how uncertainty and risk influence runtime adaptation decisions in self-adaptive systems. It synthesizes foundational concepts, canonical reference models (MAPE-K, FORMS, DYNAMICO), and diverse decision-making approaches that incorporate estimated benefit, cost, and risk. The authors argue for a unified decision-making perspective and advocate an architectural viewpoint that explicitly handles benefit, cost, and risk to guide autonomous adaptation. The work aims to facilitate robust, risk-aware self-adaptation in domains like IoT and cyber-physical systems by aligning design-time models with runtime decision processes.
Abstract
In this essay, we introduce the basic concepts necessary to lay out the foundation for our PhD research on uncertainty and risk-aware adaptation, and discuss relevant related research.